論客語的否定詞與疑問句

全文

(1)國立臺灣師範大學英語學系碩. 士. 論文. Master Thesis Department of English National Taiwan Normal University. 論客語的否定詞與疑問句. On Negation and Interrogatives in Hakka. 指導教授：吳曉虹 Advisor: Dr. Hsiao-hung Iris Wu 研究生：謝雁茹中華民國一零二年七月 July, 2013.

(2) 摘要本研究主要著重在兩種句末否定詞的問句(VP-NEG questions)，分別是當句末否定詞為 mo 或是 mang 的句構，藉由呈現客語 VP-mo 問句和中文正反問句的相同處(短語答句、副詞、語氣詞、能產度、中性問句)，我主張客語的 VP-mo 問句其實就是我們所謂的正反問句，有鑒於這些相同處，我承接了黃(1991)在對中文正反問句的分析時所提出疑問曲折詞組(Interrogative INFL)的抽象概念，我認為 VP-mo 問句中的 mo 其實就是疑問曲折詞組的體現。在這樣的分析之下，我提出客語 VP-mo 問句的形成歷經了兩個階段，首先，否定詞 mo 歷經了 NEG-I-C 的位移，再者 IP 會整個移位到 CP spcifier 而得到疑問的語意。這樣的分析也適用於分析 VP-mang 問句的形成，雖然 VP-mang 不屬於正反問句，但由於 mang 自身就帶有時貌的語意，進而驅使了否定詞 mang 歷經 NEG-I-C 的位移。另外，此研究也重新檢視了前人對客語疑問句的分類，我認為前人的觀察中，有三項是需要修正的，第一，客語句末的 mo 不應等同於中文句末的’嗎’，更進一步的說客語句末 mo 疑問句不應該被歸類為語助詞問句，因為真正的語助詞問句應該要可以跟否定詞一起出現；第二，我證明客語 A-ya-m-A 的句構不該被視為正反問句，因為 A-ya-m-A 不受限於孤島約束而且它的正向結構(A)和反向結構(not-A)是可以互換的；第三，我提出[you......mo]被視為正反問句的真正原因是因為 mo 為疑問曲折詞組的體現，並不是因為 you 和 mo 分別擔任 A 和 not-A 的角色，證據來自於在[you......mo]這樣的問句中，省略 you 並不會改變疑問的語意，這點和中文的正反問句完全不同，因為省略正向結構(A) 或反向結構(not-A)都會使得原來的句子喪失疑問的語意。. 關鍵字: 客語、疑問句、否定詞、正反問句. i.

(3) Abstract. This study focuses on two kinds of VP-NEG questions in Hakka, VP-mo questions and VP-mang questions. I argue that Hakka VP-mo question is a type of A-not-A question by demonstrating the affinities between Hakka VP-mo questions and Mandarin A-not-A questions (fragment answer, adverbs, particles, productivity, and neutral questions). Granted these similarities, I claim that mo is the overt realization of the A-not-A morpheme embedded under the interrogative INFL node in the spirit of Huang’s (1991) Modularity approach. In this way, I propose that Hakka VP-mo questions are formed by two steps. First, the negator mo undergoes NEG-I-C movement. Then the remaining IP is moved to the spec of CP for forming questions. This analysis can also apply to VP-mang questions. Though VP-mang question is not a type of A-not-A question, the negator mang inherently bears aspectual meaning, which serves as a motivation of NEG-I-C movement. I also re-evaluate previous classification of Hakka interrogatives. In this study, three kinds of mistaken ideas are modified. First, I show that Hakka sentence final mo is not equal to Mandarin ma. Thus, statements with mo in the sentence final position should not be viewed as particle questions because a true particle question is compatible with negators. Second, I demonstrate that A-ya-m-A questions actually belong to alternative questions because they are free from island constraints and the order of the positive and negative element can be exchanged. Third, I discuss the inherent reason why [you......mo] construction noted in the previous literature can be treated as A-not-A questions. I demonstrate that you is not obligatory in [you......mo] construction, contrary to Mandarin A-not-A questions in which the A part and not-A part should co-exist. Absence of either A part or not-A part will convert the questions into a declarative sentence. Key words: Hakka, Interrogatives, Negation, A-not-A questions. ii.

(4) ACKNOWLEDGEMENTS Without the support, patience and guidance of the following people, this thesis would not have been completed. It is to them that I own my deepest gratitude. First of all, my heartfelt appreciation goes to my advisor, Prof. Hsiao-hung Iris Wu, for her guidance, caring, and patience. In the process of thesis writing, she has been extraordinarily tolerant of my crazy ideas. This thesis would have remained a dream had it not been for her constant spiritual support. Further, I am much indebted to my committee members Prof. Miao-Ling Hsieh and Prof. Chyan-an Arthur Wang. Their insightful comments and carful scrutiny help improve the quality of this thesis. Besides, my gratitude is extended to Professors who taught me in National Taiwan Normal University: Prof. Chien-Jer Charles Lin, Prof. Chun-yin Doris Chen, Prof, Gerardo Fernández-Salgueiro, Prof. Jen Ting, Prof. Jen-i Li, Prof. Joy Wu, Prof. Kwock-ping Tse, Prof. Miao-Hsia Chang, by alphabetical order. I thank them for leading me to explore the diverse aspects of linguistics and teaching me to appreciate different languages from different perspectives. I am fortune to be their student. I especially thank Prof. Chun-yin Doris Chen for teaching me the right attitude I must hold toward my life. I also want to thank my former teacher Hui-I Kung at National Changhua University of Education who introduced me to linguistics. Without her great teaching, I would not know how interesting linguistics is. Moreover, I want to thank some friends whom I met at National Taiwan Normal University: Julia Wu, Jocelyn Tsai, Carol Liao and Debbie Hsu. I am thankful for their friendship. I would particularly like to thank Debbie Hsu for always picking up my phone and pulling me out of my depression. I owe a great debt to a special friend Wei-Cherng Sam Jheng, who was always willing to help and give his best suggestion. Without his encouragement and company, I could not overcome adversity during my MA life. I wish he could always keep good faith in the process. iii.

(5) of pursing his Ph. D degree. I am waiting for the day that I can call him “Dr. Jheng!’ Also, I wish to thank my best friends: Evenly Chen, Joe Pan, Karrie Chiu, Rain Liu and Veronica Tsou. I thank them for bringing me so much joy when I came back to my hometown Miao-li. Special thanks go to Karrie Chiu, who was always there cheering me up and stood by me through the good times and bad times. Last but not least, I would like to thank my wonderful parents, Huan-hui Hsieh, Tsu-chi Tu, my thoughtful sister, Yen-chi Hsieh and my naughty brother, Tung-yen Hsieh, who always believe in me in the process of pursuing this MA degree. I thank them for their unconditional love and unfailing support. This thesis is undoubtedly dedicated to my beloved family.. iv.

(6) TABLE OF CONTENTS 摘要. i. ABSTRACT. ii. ACKNOWLEDGEMENTS. iii. TABLE OF CONTENTS. v. Chapter 1 Introduction 1.1 Overview. 1. 1.2 Main goals. 3. 1.3 The Language. 3. 1.4 Structure of this thesis. 5. Chapter 2 VP-NEG Questions and A-not-A Questions 2.1 Overview. 6. 2.2 The data. 6. 2.2.1 Similarities. 7. 2.2.2 Differences. 9. 2.3 Previous studies on VP-NEG questions. 12. 2.3.1 Movement Analysis: Cheng, Huang and Tang (1996). v. 13.

(7) 2.3.2 Ellipsis Analyses. 16. 2.3.2.1 Hsieh (2001). 17. 2.3.2.2 Li (2006). 19. 2.3.2.3 R. Huang (2008). 22. 2.3.3 Remarks on previous analyses. 23. 2.3.4 Summary. 25. 2.4 Affinity between VP-mo questions and A-not-A questions. 26. 2.4.1 Fragment answers. 27. 2.4.2 Co-occurrence with Adverbs. 28. 2.4.3 Co-occurrence with Particles. 29. 2.4.4 Productivity. 30. 2.4.5 Neutral questions. 32. 2.5 Proposal. 33. 2.5.1 Huang (1991): Interrogative INFL. 34. 2.5.2 Negation as a synthetic character. 38. 2.5.3 Main Claim: NEG-I-C Movement. 40. 2.5.4 Mo preferred over Kam. 43. 2.6 Negation mang. 44. 2.6.1 Syntactic distribution and Semantic properties. vi. 44.

(8) 2.6.2 Previous studies on Hakka mang. 47. 2.6.3 Deriving VP-mang questions. 48. 2.7 Summary. 49. Chapter 3 Classification of Hakka interrogatives 3.1 Overview. 51. 3.2 Previous Classification of Hakka interrogatives. 52. 3.3 Genuine Particle questions. 55. 3.4 Genuine A-not-A questions. 56. 3.4.1 A-ya-m-A≠A-not-A. 56. 3.4.2 you……mo≠A……not-A. 60. 3.5 Tone Sandhi. 62. 3.6 Summary. 65. Chapter 4 Conclusion 4.1 Summary. 66. 4.2 Remaining issues. 67. References. 69. vii.

(9) Chapter One Introduction. 1.1 Overview. Negation has been discussed in a considerable number of studies in the field of Chinese Linguistics, such as Teng (1974), Huang (1982), Tang (1994), Ernst (1995), Lin (2003) among others. Some of these studies are descriptive, while the others are more theoretically-oriented. Though various materialized analyses of Mandarin negation have been provided insightful theoretical implications, few studies, if there is, have been devoted to the syntax of negation in Hakka, one of the Chinese dialects. This current study attempts to fill this gap. A variety of Hakka negation markers have been identified in the previous studies (Kuan, 1974; Luo, 1984; Fang, 1993). Some are mono-morphemic words such as m, mo, and mang and others are bi-morphemic words, including m-si ‘not need’, mo oi’ not want’ , mang oi ‘ not want’ and mang ho’ not yet’. In this current study, I mainly focus on three mono-morphemic negators, including m, mo, and mang, which play an important role in forming Hakka questions. As shown in (1), negator m appears in alternative questions. As for mo and mang, each of them constructs a unique question type as displayed in (2a) and 1.

(10) (2b). Examples in (3a) and (3b) show that in Hakka, those three mono-morphemic negation markers all fail to appear in A-not-A questions. (1). Alternative question ngi. xiong. oi. hi. toi-bei. jahe m. xiong. you. think. want. go. Taipei. or. think. not. oi hi. want go Taipei. ‘Do you want to go to Taipei or not?’ (2). VP-NEG question a.. gi. oi. hi. hoggao. mo?. he. want. go. school. NEG. ‘Does he want to go to school?’ b.. gi. siid. fan mang?. he. eat. rice NEG. ‘Has he eaten yet?’ (3). A-not-A question a.. gi. voi. *m/*mo/*mang. voi. he. can. not/not/not. can sing. cong-go-e?. ‘Can he sing a song ?’ b.. gi. xiong. *m/*mo/*mang. xiong. a-ba?. he. miss. not/not/not. miss. father. 2. toi-bei?.

(11) ‘Does he miss his father ?’. 1.2 Main Goals. The aims of this study are threefold: (1) to propose a proper syntactic analysis of Hakka VP-NEG questions, (2) to provide an explanatory account of the unique characteristic of Hakka A-not-A questions, (3) to re-examine the previous classification of Hakka interrogatives (Chung, 2000). I attempt to achieve those three goals by conducting a comparative study between Mandarin Chinese and Hakka. Examples of other dialects will be demonstrated only when it is appropriate or relevant for supporting my argument.. 1.3 The Language. Hakka is a Chinese dialect spoken by approximately 15% people in Taiwan. It can be subdivided into five dialects according to their accents: Sixian, Hailu, Dapu, Raoping, Shaoan. This current study is primarily concerned with Sixian Hakka. Like Mandarin Chinese, Hakka has tone sandhi as well. There are six tones in Sixian Hakka. They are yinping, yangping , shangshen, qusheng ,yinru and yangru, as exemplified 3.

(12) in table (4). (4). Tone. yinping. yangping. shangshen. qusheng. yinru. yangru. Values. 24. 11. 31. 55. 2. 5. Examples. chim. ca. jien. jia. bad. chid. ‘deep’. ‘tea’. ‘cut’. ‘borrow’. ‘eight’. ‘ straight’. When yinping is followed by yinping, qusheng and yangru , a tonal change occurs from yinping to yanping.. (5) Hakka Tone Sandhi: 24 24+. 55 5. 24 11 +. 55 5. Examples1 of the tonal change are given below:. 1. Empirical Hakka data presented in this thesis come from two sources: some come from the limited literature. and the other come from my field work. In order to assure informants’ grammaticality judgment, they need to. meet the following two requirements: either over fifty years old or achieve Hakka proficiency test, 4.

(13) ziim. xien2. needle. line. Base tone. 24. 55. Tone sandi. 11. 55. mai. xin. sam3. buy. new. clothes. Base tone. 24. 24. 24. Tone sandhi. 11. 11. 24. (6). (7). 1.4 Structure of the thesis This thesis will be organized as follows. In Chapter Two, I begin with presenting some linguistic facts of Hakka negation and interrogatives. Primary attention will be paid to the negator mo in the first half of this chapter. In the remainder of this chapter I will discuss mang as well. Chapter Three is to re-evaluate the previous classification of Hakka interrogative sentences. Finally, Chapter Four will conclude this study with further implications and remaining issues spelled out. .. High-Intermediate level held by Hakka Affaire Council. 2. Transcription used in this thesis follows the Taiwan Hakka Romanization System established by the. Ministry of Education in Taiwan in 2012. 3. Example (6) and (7) are drawn from http://php.digimagic.com.tw/happyhakka/b1.php?id=2&up_id=37. 5.

(14) Chapter Two VP-NEG Questions and A-not-A Questions. 2.1 Overview. This chapter proceeds as follow: In section 2.2, I show relevant linguistic data first. In Section 2.3, previous analyses of VP-NEG questions will be introduced. The discrepancies and inadequacies of previous analyses will be pointed out. Later in section 2.4, I would offer five pieces of evidence to show the affinity between VP-mo questions and A-not-A questions. Given this affinity, my main proposal comes in section 2.5. The VP-mang questions will be examined in section 2.6. In this section, I illustrate the syntactic distribution and semantic properties of negator mang. An immediate consequence is that the proposed analysis can further account for the VP-mang questions. Section 2.7 concludes this chapter.. 2.2 The data. Generally speaking, Mandarin and Hakka share a lot of syntactic properties in common. They have the same word order and it is very often the case that a sentence in Hakka can 6.

(15) map word-by-word onto Mandarin. In what follows, I will present similarities and differences between these two dialects, serving the baseline for the ensuing discussion.. 2.2.1 Similarities. Mandarin Chinese has two negation markers, i.e. bu and meiyou. They share common syntactic distributions, namely, pre-verbal position or sentence-final one. When bu and meiyou occur in their typical preverbal position, negation can take their scope over their following verbal phrase and negate the whole event. On the contrary, when it is placed in the sentence-final position, it can still take its scope over the clause, and convert a declarative into an interrogative. Sentences in (1) and (2) exemplify the descriptions above. (1) a.. ta. meiyou. mai. na. jian yifu.. she. NEG. buy. that CL clothes. ‘She didn’t buy that clothes.’ b.. ta. mai. na. she buy. jian. that CL. yifu. meiyou?. clothes. NEG. ‘Did she buy that clothes?’ (2) a.. ta. bu. hui. qu. she. NEG. will go. xuexiao. school 7.

(16) ‘She will not go to school.’ b.. ta. hui qu. she will go. xuexiao bu? school. NEG?. ‘Will she go to school?’’ As for Hakka, there are three mono-morphemic negators, i.e., m, mo and mang. A slight difference between these three negation markers is that not all of them can stay in the sentence final position. Example (3) and (4) show that mo and mang can either precede a verb or locate in the end of the sentence whereas m is only permitted in pre-verbal position, as presented in (5). (3) a.. gi. mo. oi. hi. hoggao.. he. NEG. want. go. school.. ‘He doesn’t want to go to school.’ b.. gi. oi. hi. hoggao. mo?. he. want. go. school. NEG. ‘Does he want to go to school?’ (4) a.. gi. mang. siid. fan.. he NEG. eat. rice. ‘he hasn’t eaten yet.’ b.. gi. siid. fan mang? 8.

(17) he. eat. rice NEG. ‘Has he eaten yet?’ (5) a.. gi. m. voi. sii. cha-e.. he. not. can. derive. car. m?. ‘He can’t drive a car’ b.. *gi. voi. sii. cha-e. he. can. drive. car. NEG. ‘Can he drive a car?’ At first glance, it predicts that the Hakka VP-NEG questions can be analyzed on a par with the Mandarin VP-NEG questions in a principled manner. This prediction, as a matter of fact, is not carried out and is countered by the evidence in the next section, which therefore triggers a need to propose a proper analysis of the Hakka VP-NEG questions.. 2.2.2 Differences. An interesting contrast between Mandarin Chinese and Hakka is that verbs, adjectives, modals, adverbs and prepositions used in deriving Mandarin Chinese A-not-A questions fail to form A-not-A questions in Hakka, as illustrated from (6) to (10).. 9.

(18) (6) Verb a.. ni. he. you drink. jiu. bu. he. jiu?. wine. not. drink. wine. ‘Would you like to drink?’ b.. *ngi. siid. you. drink. jiu. m. siid. jiu?. wine. not. drink. wine. ‘Would you like to drink?’ (7) Adjective a.. zhe ge this. nu-sheng CL. girl. piao-liang. bu. piao-liang?. beautiful. not. beautiful. ‘Is this girl beautiful?’ b.. * ia. ge. sia-moi-e. this. CL girl. jiang. m. beautiful not. jiang? beautiful. ‘Is this girl beautiful?’ (8) Modal a.. ta. neng. he can. bu. neng. not can. he. jiu?. drink wine. ‘Can he drink wine?’ b.. *gi zo-ded. bu. zo-ded. siid 10. jiu?.

(19) he. can. not. can. drink. wine. ‘Can he drink wine?’ (9) Adverb a.. ta. chang. bu. she. often. not. chang often. mai. yi -fu?. buy. clothes. ‘Does she often buy new clothes?’ b.. *gi. jiab. m. jiab. mai. sam-fu?. she. often. not. often. buy. clothes. ‘Does she often buy new clotes? ‘ (10) Preposition a.. ta. jin-tian. zai. bu. zai. he. today. at. not at. jia? home. ‘Is he at home today?’ b.. *gi. jin-pu-ngid. di. m. di. vug-ga?. he. today. at. not. at. home. ‘Is he at home today?’ Over the past decades, a lot of works have investigated A-not-A questions. Issues discussed in these works range from the descriptions of different sub-types of this construction, its correlation with alternative questions and WH-questions, to its (in)compatibility with 11.

(20) different adverbials and focus makers(Wang, 1967; Lin, 1974; Huang, 1982, 1991; Kuo, 1992; Zhang 1997; Hsieh, 2001; Gasde, 2004; Huang, 2008; Tseng, 2009). However, the aforementioned phenomenon has not been noted yet. The differences in surfacing A-not-A questions between Mandarin Chinese and Hakka will receive a proper account in the end of this chapter.. 2.3 Previous studies on VP-NEG questions. In section 2.2.1, I have shown that Hakka negators have similar syntactic distribution with Mandarin negators. In these two dialects, negator can serve as a sentential negation or forming a VP-NEG question when it is placed in the sentence final position. Thus, in this section, I would first review previous analyses on VP-NEG questions and then see whether previous analyses can account for the abovementioned Hakka data. The debate primarily centers around whether VP-NEG questions are reduced from A-not-A questions or not. The ellipsis hypothesis is defended in Hsieh (2001), Li (2006) and Huang (2008) whereas Cheng, Huang and Tang (1996) argue for the movement analysis. I will present their analyses in the following.. 12.

(21) 2.3.1 Movement Analysis. With regard to VP-NEG4 questions in Mandarin, Taiwanese and Cantonese, Cheng, Huang and Tang(1996) observe a distinction between these three dialects; that is, Mandarin VP-NEG questions displays a selection relation between the negation marker and the aspect/verb, whereas Cantonese and Taiwanese do not. All of these three dialects have more than one negation marker. In Mandarin Chinese, bu is used with bare verbs and modals, and meiyou is used with various aspects and with accomplishment verbs. As shown in (11) and (12), meiyou cannot appear with the modal verb hui, and bu can not appear with the aspect marker -guo. The selection relation maintains in VP-NEG questions as well. (11). a.. huifei. bu/*meiyou. hui. qu.. Huifei. NEG/NEG-have. will. go. ‘Huifei will not go.’ b.. huifei Huifei. *bu/meiyou. qu-guo.. NEG/NEG-have. go-Asp. ‘Huifei has not been (there).’ 4. Cheng, Huang, and Tang (1996) named this construction as Negative Particle Question (NPQ). Since in this. study, we do not treat sentence final negation as a particle, we will not use the term NPQ but consistently use VP-NEG questions throughout this thesis. 13.

(22) (12). a.. ta. hui. qu. he. will. go. bu/*meiyou? NEG/NEG-have.. ‘Will he go?’ b.. ta. qu-guo. *bu/meiyou?. he. go-asp. NEG/NET-have. ‘Has he been (there)? ’ As for Cantonese, there are three negation markers, i e., m, mou, and mei. Cheng, Huang and Tang mention that m is on a par with bu in Mandarin. It appears with bare verbs and modal. Mou behaves like Mandarin meiyou, which is used with various aspects and accomplishment verbs. Mei is similar to mou except for the fact that the former has an extra meaning of ‘not yet’. Unlike Mandarin VP-NEG questions, which allow bu and meiyou to be used in VP-NEG questions, mei is the only negation marker that can appear in VP-NEG questions in Cantonese. Presumably, judging from example (13), it is expected that VP-NEG questions can not contain modals since mei cannot appear with modals in declarative sentences. However, it turns out that mei can still appear with modal (14a), perfective(14b) and experiential aspect(14c) in VP-NEG questions. In other words, the selection restriction between negation and aspect/verb does not need to be maintained in Cantonese VP-NEG questions.. 14.

(23) (13). keoi. m/*mou/*mei. hoyi. lei.. he. no/not-have/not-yet. can. come. ‘He will not come.’ (14). a.. ngo. hoyi/yinggoi ceot-heoi. *m/*mou/mei?. I. can/should. go-out. NEG/NEG-have/NEG-yet. ‘Can/should I go out?’ b.. keoi. heoi-go. meigwo. *m/*mei/mei?. he. go-EXP. America. NEG/NEG-have/NEG-yet. ‘Has he been to America?’ c.. keoi. sik-zo. fan. mei?. he. eat-PERF. rice. not. ‘Has he eaten?’ Taiwanese is similar to Cantonese in showing no strict requirement for a selection between negative forms and aspect/verb in VP-NEG questions. In (15), it is shown that m, bo and buei can not occur with the modal e “will” in declarative sentence. Nevertheless, in VP-NEG questions, all the negation makers can appear with e “will”. (15). a.. *i. m/bo/buei. e. lai.. he. not/not-have/not-yet. will. come. ‘He will not come’ 15.

(24) b.. i. e. lai. he. will. come. m/bo/buei/be? not/not-have/not-yet/not-FUT. ‘Will he come?’ In order to account for the selection relation between the negation markers and the aspect/verb, Cheng, Huang and Tang propose that Mandarin VP-NEG questions are derived from NEG-to-C movement. Following Chomsky (1995), they assume that NEG-to-C movement is triggered by the unchecked feature. In light of feature checking, they assume that there exists a phonologically null C0 with formal features [Q, NEG]. Negators bu or meiyou located in Neg0 will move up to C0 for checking [ NEG] feature. As for Taiwanese and Cantonese VP-NEG questions, they argue that negation markers used in VP-NEG questions are base-generated in C0. Those negation makers have “dual status” which can serve as negation markers or as question particles.. 2.3.2 Ellipsis Analyses. With regard to VP-NEG question, three pieces of works, which all stand for the ellipsis analyses, will be reviewed in this section: Hsieh (2001), Li (2006) and R. Huang (2008). Though they all claims that VP-NEG questions are formed by eliding the constituents followed by the negators, their assumption on the underlying structure of VP-NEG 16.

(25) questions is different. Hsieh and R. Huang both argue for that VP-NEG questions derives from VP-not-VP questions via deletion of the second occurrence of VP. As for the underlying structure of VP-not-VP questions, Hsieh proposed that [VP] and [not VP] are connected with a coordinated &P, but R. Huang argues that VP-not-VP derives from a reduplicated rule. Unlike Hsieh and R. Huang, Li claims that VP-NEG questions are derived from the base structure of juxtaposed IPs. More details will be presented in the following subsection.. 2.3.2.1 Hsieh (2001). Hsieh argues that meiyou heads a NegP, which occurs between the aspect phrase and predicate phrase. For another negation m, it is argued to be adjoined to V bar or predicate bar. She further distinguishes VP-bu questions from VP-meiyou questions by showing that meiyou can license a null VP, while bu cannot. Consider the contrast below. (16) a.. Zhangsan. mei-le. nei-ben. shu, keshi. Lisi. meiyou [vp e]. Zhangsan. buy-PERF. that-CL book but. Lisi. not have. keshi. Lisi. bu [vp e]. Lisi. not. ‘Zhangsan bought the book, but Lisi didn’t.’’ b.. *Zhangsan. xihuan. zhe-ben shu,. Zhangsan. like this-CL book but 17.

(26) ‘Zhangsan likes this book, but Lisi doesn’t.’ With this contrast in mind, it should be noted that for Hsieh, only VP-meiyou questions can be regarded as a true VP-NEG questions in Mandarin Chinese. Hsieh attempts to propose an unified analysis for VP-NEG questions and A-not-A questions. Following Aoun and Li (1993), she proposes that in Mandarin Chinese, there exists a QP (question phrase) and an operator in the specifier of QP. The representation for a VP-NEG questions is depicted below. (17) CP 3 C’ 3 QP 3 Operator. C. Q’ 3. XP 3 XP 4. Q [+WH]. XP 4. (Hsieh, 2001:128). VP-NEG questions have the [+WH] feature base-generated under the functional head Q, and that the [+Q] operator moves from the [Spec, QP] to [Spec, CP] at LF in order to check the strong [+Q] feature of C head. In light of Progovac (1999), Hsieh further proposes a more elaborate structure, which 18.

(27) has advantages over the above simple structure in analyzing VP-NEG questions. A coordinate &P connects the two constituent, namely [VP] and [not VP], and the & head is presumably a null form in VP-NEG questions. With these assumptions, she proposes that VP-NEG questions contain a QP taking a coordinate construction with VP ellipsis on the second conjunct. Consider a VP-NEG question in (18a) . Example (18b) demonstrate its simplified structure. (18) a.. ta. kan-le. nei-ben. shu. meiyou [vp e]?. he. read-PERF. that-CL. book. not. [vp e]. ‘Did he read that book or not.’ b. [CP Opi [TP Ta [ QP ti Q [+WH] [ &P [VP kan-le na-ben shu ]…&…[NEGP meiyou [ VP e]]]]]] As shown in example (18b), kan-le na-ben shu ‘read-asp that book’ [VP] and meiyou kan-le na-ben shu ‘not-have read-asp that book’ [not-VP] are connected by &P. Then example (18a) is derived via deleting the verbal phrase following the negator meiyou.. 2.3.2.2 Li (2006). In dealing with VP-NEG questions, Li first argues against movement analysis by showing that it not only rules out grammatical sentences, but also generate ungrammatical 19.

(28) sentences. Example (19a) shows that in declarative sentences, both bu and meiyou can not occur with the perfective aspect marker le, but in VP-NEG questions, meiyou can occur with le. The movement analysis fails to explain the contrast between (19a) and (19b). If it is assumed that meiyou is extracted from the inside IP, we would expect a ungrammatical sentence. However, example (19b) shows that when negator meiyou is in the sentence-final position, it is able to co-exist with aspect -le. (19)a.. *ta. bu/meiyou. qu-le. xuexiao.. he. NEG/NEG-have. go-PERF. school. ‘He didn’t go to school.’ b.. ta. qu-le. xuexiao. bu*/meiyou?. he. go-PERF. school. NEG/NEG-have. ‘Did he go to school?’ Along the same line, if we believe VP-NEG questions are formed by moving the negation maker to C0, we would expect that example (20b) to be grammatical, which is again not the case here. (20)a.. ta. bu. zhi xihuan. zhe ben shu.. he. NEG. only like. this CL book. ‘He does not only like this book.’ b.. *ta zhi. xihuan. zhe. ben shu 20. bu?.

(29) He only. like. this. CL book. NEG. ‘Does he only like this book?’ The abovementioned examples illustrate that the movement analysis fails to make correct predictions. Given this, Li suggests that VP-NEG questions are derived from the base structure of juxtaposed IPs,5 which undergo anaphoric ellipsis. The constituent that follows the negation marker will be deleted. Her analysis not only explain the data in Mandarin but also in Cantonese and Wenzhou. Take Mandarin VP-bu and VP-meiyou questions for illustration. The derivation of (21a) is demonstrated in (21b). (21) a.. ta. hui. qu. xuexiao bu?. he. will. go. school. NEG. ‘Will he go to school? b.. ta[IP1 [VP hui qu xue xiao]] [IP2 [NEGP[bu[VP hui qu xuexiao]]] ta[IP1[VP hui qu xiexiao] [IP2 [NEGVP[bu[VP hui qu xueuxiao]]] ta[IP1 [VP hui xu xuexiao]] [IP2 [NEGP[bu]]. Example (21b) shows that the reason why negation marker bu appears in sentence-final is due to the deletion of the constituent followed by it. As for VP-meiyou questions, for Li, meiyou is analyzed as a combination of the 5. Slightly differ from Huang (1991) who suggests that the base structure of VP-NEG question consists of. juxtaposed VPs,. Li (2006) assumes that they are IPs, because the conjuncts may comprise aspect marker and. modals. 21.

(30) negation maker mei and the verb you. Therefore, deletion of the complement of the verb you will lead to a question ending with a negation word ,as demonstrated in (22b). (22). a.. ta. you. qian. meiyou?. he. have. money. NEG-have. ‘Does he have money?’ b.. Ta[ IP1[VP you [NP qian]]] [IP2[NEGP me i[VP you[NP qian]]] ta[IP1[VP you [NP qian]]] [IP2 [NEGP mei [VP you [NP qian]]] ta[IP1[VP you [NP qian]]] [IP2 [NEGP mei [VP you]]]. 2.3.2.3 R. Huang (2008). In his work, R. Huang proposes an unified analysis of deriving various types of A-not-A questions. Under his proposed analysis, VP-NEG question is a subtype of A-not-A questions. I list all types of A-not-A questions below first and then explain the derivation of VP-NEG questions under R. Huang’s proposal later. (23) a. ta he. xihuan. zhe-ben shu bu. xihuan. like. this-CL book not like. zhe-ben shu?. (VO-not-VO). the-CL book. ‘Does he like this book or not like this book?’ b. ta. xihuan. zhe-ben shu. bu 22. xihuan?. (VO-not-V).

(31) he. like. this-CL book. not like. ‘Does he like this book or not like(it)?’ c. ta he. xihuan. bu. xihuan. like. not. like. zhe-ben. shu?. (V-not-VO). this-CL book. ‘Does he like or not like this book?’ d. ta he. xi-bu-xihuan. zhe-ben shu?. li-not-li-ke. this-CL book. (V-not-VO). ‘Does he li-or no like this book?’ e. ta he. xihuan. zhe-ben shu. bu ?. like. this-CL book. not. (VO-not). ‘Does he like this book or not?’ He pointed out that under Huang’s modular approach (1991), VO-not-VO questions like (23a) can actually be derived in two ways: either from a base-generated coordinate [VO] [not VO] structure or from a simplex structure which undergoes reduplication. He adopts the latter approach. In this account, all types of A-not-A questions can be derived in a unified way. In dealing with different types of A-not-A questions, all he needs to assume is the existence of the abstract Interrogative INFL and the process of anaphoric analysis. Examples (23a), (23c), and (23d) are derived from a reduplications rule, because their reduplicated elements are all “phonological unit”. Example (23c) selects a verb xihuan ‘like’ 23.

(32) to be the reduplicated element while example (23a) chooses the whole verbal phrase xihuan zhe ben shu ‘ like this book’ to reduplicate. In the matter of VO-not-V (23b) and VP-NEG (23e), R. Huang argued that they are derived by applying the anaphoric ellipsis to a reduplicated VO-not-VO structure.. 2.3.3 Remarks on previous analyses. Each of the works reviewed in the preceding section will not be commented one by one in my study because most of them face the same problem in accounting for the data observed in Hakka. Therefore, I will demonstrate that Hakka VP-NEG questions cannot receive a proper analysis under either movement analysis or ellipsis analysis. Let us consider the ellipsis analysis first. If we agree example (24) is derived from ellipsis, this means that the original sentence will look like (25). However, it is shown that the presumably pre-deleted constituent voi hi honggao ’will go to school’ cannot be negated by mo; instead, it should be negated by another negator m as demonstrated in (26). (24) gi. voi hi. hoggau. mo?. he. will go. school. NEG. ‘Will he go to school?’. 24.

(33) (25) *gi voi hi he will go. hoggau. mo. voi. hi. hoggao?. school. NEG. will. go. school. ‘Will he go to school?’ Likewise, if we assume that after example (26) undergoes ellipsis, a VP-NEG question will be derived. Nevertheless, when the post-negator constituent voi hi hongao ‘will go school’ is elided, the ellipsis process renders the sentence ungrammatical, as illustrated in (27). The ungrammaticality of (27) is because pure negation m is not able to appear in the sentence final position. These ungrammatical sentences amount to the fact that ellipsis analysis fails to make correct predictions in Hakka. (26) gi. voi hi. he will go. hoggau. m. voi. school. NEG. will. hi go. hoggao? school. ‘Will he go to school?’ (27) *gi he. voi hi. hoggau. m?. will go. school. NEG. ‘Will he go to school?’ Next, if we concur the Neg-to-C movement analysis, we would have difficulty in explaining example (28), where the question particles no or li has already occupied the head position of C. Under this circumstance, in accounting example (28), I claim that spit C analysis is needed. Moreover, though I agree that the final landing site of the negator is the head of C, I 25.

(34) argue that Hakka negator does not move directly from the preverbal position to the head of C. I would demonstrate that Hakka negators would undergo NEG-I-C movement in forming VP-NEG questions in this study. (28) gi. voi hi. hoggau. mo. no/li?. he. will go. school. NEG. particle?. ‘Will he go to school?’. 2.3.4 Summary. Analysis. Movement. Base-generated. Ellipsis. C.H.T.(1996). Hsieh(2001) Li(2006) R.Huang(2008) Li(2006). Dialects Mandarin Chinese. Cantonese. C.H.T.(1996). Li(2006). Wenzhou Taiwanese Southern Min. C.H.T.(1996). Hsieh(2001). The above table summarizes the previous analyses. With regard to VP-NEG questions, different Chinese dialects have been investigated including Mandarin Chinese, Cantonese, Wenzhou and Taiwanese southern Min. Some argue for the NEG-to-C movement and the others argue for the ellipsis analysis. However, neither of them can account for Hakka 26.

(35) VP-NEG questions. Given that, it is necessary to examine Hakka VP-NEG questions.. 2.4 Affinity between VP-mo questions and A-not-A questions. A-not-A questions have a lot of noteworthy syntactic and pragmatic characteristics. A number of diagnoses have been developed to determine whether a particular question is a A-not-A question or not. In the following section, I will show that the characteristics observed in Mandarin A-not-A questions can also be found in Hakka VP-mo questions. Mandarin dada will be illustrated first and then Hakka.. 2.4.1 Fragment answers. As pointed out by Hagstrom (2006:174), a major difference between A-not-A questions and yes/no questions is that A-not-A questions can not be answered with shi ’yes’ dui ’correct’ or bu ’no’; instead they should be answered with the positive or negative form of the main verb. As shown in example (29b), A-not-A questions are answered by repeating either the questioned predicate ai ‘love’ or its negation bu ai ‘ not love’. (29) a.. A:. xiao-ming. ai-bu-ai. chi qiao-ke-li?. John. love-not-love. eat chocolate. 27.

(36) ‘Does John love to eat chocolate?’ B:. ai/ bu ai/ *shi/ *bu / *dui love/ not love/ yes/ no/ correct. As evidenced below, the way we answer VP-mo questions are of no difference. Example (31) shows that the short answer form of the VP-mo questions can’t be “Yes” or “No”. The correct short answer form should be the positive or negative of the main verb. (30)A:. gi. voi hi. hoggau. mo?. he. wll go. school. NEG. ‘Will he go to school?’ (31) B:. voi/. m. voi /*hei. will/. not will/ Yes. /*m hei/ *dui / not yes/ correct. 2.4.2 Co-occurrence with adverbs. Second, it is well accepted that A-not-A questions can’t occur with nandao’actually’ whereas they are compatible with daodi’ truly’. Consider example (32) (32). ta. daodi/*nandao. yao-bu-yao. he. truly/*actually. want-not-want. lai. gen women. chi wancan?. come. with us. eat dinner. ‘Let me go the truth: does he want to have a dinner with us?’ 28.

(37) Example (33) demonstrates that like A-not-A questions in Mandarin Chinese the VP mo questions are incompatible with the adverbs nandao‘actually’. (33). gi. dodi/ *mosheng. voi. hi. hoggau. mo?. he. truly /actually. will. go. school. NEG. ‘Let me go to the answer now: will he go to school ?’ As can be seen above, VP-mo questions and Mandarin A-not-A questions behave similarly. Both of them are able to co-exist with the adverbial daodi ’truly’ but not with nandao ’actually’.. 2.4.3 Co-occurrence with particles. Particles always attach to declaratives. Some are put in the sentential initial position, others are located in sentence medial, still other are positioned in the sentence-final position. It is observed that in Mandarin Chinese (Cheng, Huang, Tang, 1996), particles can be added to the end of A-not-A questions, as shown in (34) and (35). (34) ni. xihuan-bu-xihuan zhe ge. you like-not-like. this CL. nv-sheng. ne?. girl. particle. ‘Do you like this girl or you don’t like this girl?’ (35) ni. xihuan-bu-xihuan zhe. ge nv-sheng 29. la?.

(38) you like-not-like. this CL girl. particle. ‘Do you like this girl or you don’t like this girl?’ Concerning Hakka VP-mo questions, different particles can appear in the end of them as well. (36). gi he. voi hi. hoggau. mo. no?. wll go. school. NEG. particle. ‘Will he go to school?’ (37). gi. zhong-i ya. he like. ge. this CL. sia-moi-e. mo. li?. girl. NEG. particle. ‘Does he like this girl?. 2.4.4Productivity. In Mandarin Chinese, in forming A-not-A questions, almost every syntactic category can fit into the slot in the A-not-A formulation. I show that Hakka mo can attach to the end of any declarative despite the syntactic category of the main predicate. (38)Verb a.. ni. he. you drink. bu. he. jiu?. not. drink. wine 30.

(39) ‘Do you drink?’ b.. ngi. siid. jiu. mo?. you. drink. wine. NEG. ‘Do you drink?’ (39)Adjective a.. zhe. ge. nu-sheng. piao-liang. bu. piao-liang ?. this. CL. girl. beautiful. not. beautiful. sia-moi-e. jiang. mo?. girl. beautiful. NEG. ‘Is this girl beautiful?’ b.. ia this. ge CL. ‘Is this girl beautiful?’ (40)Modal a.. ta. neng. bu. neng. he. can. not can. he. jiu?. drink. wine?. ‘Can he drink wine?’ b.. gi. zoded. siid. jiu. he. can. drink wine. mo? NEG?. ‘Can he drink wine?’. 31.

(40) (41)Adverb a.. ta chang. bu. chang. she often. not often. mai. yifu?. buy. clothes?. ‘Does she often buy new clothes?’ b.. gi. jiab. mai. sam-fu. mo?. she often. buy. clothes. NEG. ‘Does she often buy new clothes?’ (42)Preposition a.. ta jin-tian. zai. bu. zai jia?. at. not. at. he today. home. ‘Is he at home today?’ b.. gi. jin-pu-ngid. ti. vug-ga. mo?. he. today. at. home. NEG. ‘Is he at home today?’. 2.4.5 Neural questions. Last, I show that like A-not-A questions, VP-mo questions are used in neutral contexts where the speaker has no presupposition about the answer. Consider the following situation: 32.

(41) Judy’s brother brings his new girlfriend named Karen home. He introduces Karen to his grandmother. Judy observes the interaction between Karen and her grandmother secretly. When she hears her grandmother laughs loudly, she thinks that her grandmother must really like Karen. With this assumption in mind, all she wants to do is to confirm this assumption with her brother. In this given context, only (43b) is felicitous. Example (43a) is not felicitous because it is a neutral questions indicating that the positive answer and the negative answer are equally unknown to the speaker. (43) a. # apo grandma. zhong-i ya-ge like. se-moi-e mo?. this-CL girl. NEG?. ‘Does grandmother like this girl?’ b.. apo. zhong-i ya-ge. grandma. like. se-moi-e ho?. this-CL girl. particle?. ‘Does grandmother like this girl?’ Intended meaning: ‘I have already known that grandmother really liked the girl because I heard they chatted happily in the living room.’. 33.

(42) 2.5 Proposal. After providing five pieces of evidence to show the affinity between VP-mo questions in Hakka and the A-not-A questions in Mandarin Chinese, I would like to review Huang’s (1991) work on analyzing A-not-A questions for the purpose of being a preface to my main claim. In addition to that, I adopt Li’s (1971) assumption on treating negation markers as a synthetic word. This assumption is essential for serving as a motivation for NEG-to-I movement proposed in this study. More details of this assumption will be presented in section 2.5.2.. 2.5.1 Interrogative INFL: Huang (1991). Wang (1967) analyzes A-not-A questions and Alternative questions in a unified way. As opposed to this one-rule approach, Huang(1991) argued that A-not-A questions differ from disjunctive questions because these two types of questions behave differently with respect to lexical integrity, preposition stranding and island constraints. Moreover, he further stated that A-not-A questions should be separated into two subtypes and suggested that each of them is derived in different ways. In Huang’s analysis, a question like example (44) is regarded as AB-not-A questions. 34.

(43) This AB-not-A form is derived through Anaphoric Ellipsis from the base-generated coordinate form. The second occurrence na-ben shu ‘that book’ will be deleted via anaphoric ellipsis. (44) ta. xihuan. He like. naben. shu. that-CL book. bu. xihuan naben. not like. shu?. that-CL book. ‘Does he like that book (or) doesn’t he like that book?’ Moreover, it is mentioned by Huang that the anaphoric ellipsis has to meet the following two requirements. First of all, the process of anaphoric ellipsis can not destroy lexical integrity. Definition of lexical integrity hypothesis is given below. (45) Lexical Integrity Hypothesis (Huang, 1991:310) Phrase-level rules belonging to the syntactic component cannot affect a proper sub-part of a lexical category (word). Example (46b) is ungrammatical because it violates Lexical Integrity Hypothesis, according to which the lexical verb xihuan ’like’ is broken by the process of anaphoric ellipsis. Another requirement is that anaphoric ellipsis should not be in conflict with preposition stranding. Ungrammatical cases that violate preposition stranding is illustrate in (47b). (46). a.. ta. xihuan. zhe-ben shu. he. like. this-CL book. bu not. xihuan? like. ‘Does he like this book (or) doesn’t he like this book’ 35.

(44) (47). b.. *ta xihuan zhe-ben shu bu xi-?’. a.. ni. cong. zheli. chu-qu. bu. cong. zheli. chu-qu?. you. from. here. go-out. not from. here. go-out. ‘Will you from here go out (or) not form here go out?’ b.. *ni cong. zheli. chu-qu. bu. [PP cong [e] ]?. you from. here. go-out. not. from. ‘Will you form here go out(or) not form (here go out)?’ Differing from AB-not-A questions, A-not-AB questions are derived through reduplication from a simplex sentence with a abstract Q under INFL, as represented below. (48). (Huang, 1991:316) In Mandarin Chinese, this abstract Q is realized by a process of reduplication. Meanwhile, negation will be inserted into between the original and the copy element. The reduplicants length can range from a syllable (or morpheme) to the entire VP which immediately follows the interrogative functional head. Therefore, the result will be xi-bu-xi 36.

(45) ‘li-not-like’ (49a), xihuan-bu-xihuan ‘like-not-like’ (49b), xihuan zheben shu bu xihuan zheben shu’ like this book not like this book (49c). (49). a. Ni xi-bu xihuan zhe-ben shu? b. Ni xihuan bu xihuan zhe-ben shu? c. Ni xihuan zhe-ben shu bu xihuan zhe-ben shu? ‘Do you like this book (or) not like this book?. Another statement worth mentioning is that the reduplicant should be a “phonological unit”. Since Huang (1991) did not give a clear definition of what he calls a “phonological unit’, we agree with R. Huang ‘s (2010) idea in which the phonological unit is syntactic constituent. Therefore, as evidenced below, since xihuan zhe ‘like this’ can not be viewed as a syntactic constituent which results in an ungrammatical sentence. (50) *ni xihuan you like. zhe bu this. xihuan zhe. not like this. shu? book. “Do you like this (or) not like this book?’ In order to lend support to the existence of an abstract Q, Huang (1991) argues that Taiwanese kam-question is a variant of A-not-A questions. It is stated that the abstract Q can not only be phonologically realized by reduplication but also can be lexically realized in other Chinese dialects. Examples (51) to (53) show the realization of the abstract Q in Early Mandarin (ke), Taiwanese (kam i) and in Shaghainese (e) respectively. 37.

(46) (51). li. you. kam Q. beh want. lai? come?. ‘Do you want to come?’ (52) nong you. a. le. va?. Q. come. Particle?. ‘Will you come?’ (53) ni. ke. you Q. you. kong?. have. time?. ‘Do you have time?’. 2.5.2 Negation as a synthetic character. Li (1971:207) distinguishes four negative markers in Southern Min: bo ’not have’, be ’unable’, volitional m1 ‘not want, will not’ and non-volitional m26 ’not’. According to him, he argues that m2 is a pure negation marker and all the other negative markers are verbs derived from their corresponding affirmatives. (54) a. 6. NEG + u’have’ bo ’not-have’. There has long been discussion on whether there is only one m or two different m’s, volitional m and simple. m, in Taiwanese. Readers interested in this issues can refer to Lin, 1974; Cheng, 1997; Teng, 1992, Tang, 1994; Yang, 2004. 38.

(47) b.. NEG + e’able’ be ‘not-able’. c.. NEG + beh m1 ‘not-want’. d.. NEG + Ø  m2 ‘not’. Linguists (Teng, 1992; Lin, 2004; Yang, 2012) who devoted to the research of negation in Taiwanese Southern Min (TSM) agree with Li’s viewpoint that TSM negative markers are a phonetic fusion of m/NEG with another morpheme. Here, in dealing with Hakka negation, I mainly follow Li’s viewpoint on assuming that the mono-syllabic negation is comprised of two morphemes in its underlying structure. Here, I assume that Hakka mo consists of a pure negation and its affirmative counterpart you, whereas mang is comprised of a pure negation and an aspect morpheme e. (55). a. m+ aspect you. mo. b. m+ aspect e7 mang c. m+ Ø  m (pure negation) (56) a.. 7. ni. chi. bao. le. mei?. (Mandarin). b.. *ngi. siid. bao. e. mang?. (Hakka). c.. ngi. siid. bao. mang?. (Hakka). You. eat. full. aspect. NEG?. Yeh (2001) argues that Mandarin –le and the Hakka ∅ marker are corresponding forms. Our assumption. also supports her viewpoint in whish she claimed that Hakka aspect marker -le is not lexically realized. 39.

(48) ‘Have you eaten yet?’ Example (56) demonstrates that mang cannot co-exist with aspect e. Given that, the negation mang has already contained an aspect maker e in its underlying structure. Thus, the same aspect makers in one single statement renders a sentence ungrammatical. Another piece of evidence comes from the comparison between (a) and (c). It is quite clearly that Mandarin le mei corresponds to one single word in Hakka which is phonetically realized as mang.. 2.5.3 Main Claim: NEG-I-C Movement. Along the line of showing the affinity between the VP-mo questions and real A-not-A questions in section 2.4, I argue that Hakka sentence-final mo is the overt realization of the A-not-A morpheme embedded under the interrogative INFL in the spirit of Huang’s (1991) Modularity approach. Note that mo is not base-generated under this interrogative INFL. When it serves as a sentence negation, it is base-generated under the head of NEG, while it moves from Neg to I and to C in forming questions. The following tree diagram illustrates how (57) is derived. Negation mo first undergoes head-to-head movement. Then the remaining IP will move to the specifier of C in forming a question.. 40.

(49) (57). ngi. siid. jiu. mo?. you. drink. wine. NEG. ‘Do you drink?’ (58) Step one: NEG-I-C movement CP 3 C’ 3 C IP moi 3 gi. I’ 3 I ti. NegP 3 Neg’ 3 Neg ti. VP 5 siid jiu. (59) Step two: Move the remaining IP to spec of C CP 3 IPj 6 gi. siid. C’ 3 C moi. jiu. tj. Next, I will demonstrate the way to derive a VP-NEG question with a particle attached to it. Mo still undergoes NEG-I-C movement. During the process of movement, mo brings 41.

(50) the particle no located in the head of CP1 up to the head of CP2. Then, the remaining IP will move to the specifier of CP1 and then to CP2.. (60) Step one: NEG-I-C1 movement CP2 3 C’ 3 C2 moi-noj. CP1 3 C’ 3 C1 tj. IP 3 gi. I’ 3 I ti. NegP 3 Neg’ 3 Neg ti. (61) Step two: move the remaining IP to spec of C1 and then to C2 CP2 3 IPK 4 gi. siid. jiu. C’ 3 C2 moi-noj. CP1 3 tk. C’ 3 C1 tj 42. tk. VP.

(51) 2.5.4 Mo8 preferred over Kam. In conformity with Huang(1991), our analysis supports his proposal of interrogative INFL in analyzing Mandarin A-not-A questions. Contrary to Huang who argues that interrogative INFL can be lexically realized in Taiwanese kam, we would like to show that it is more convincing to argue that this abstract element is incarnated by mo in Hakka for two reasons. Theoretically, an overt or covet element should yield the same semantic meaning. Chinese A-not-A questions are a type of question without any presupposition. The speaker does not have the answer in mind in advance and the listener will answer the question based on his/her world knowledge but not according to some hints, such as intonation change from the speaker. Note that in Taiwanese, statements with kam are not a neural question. When the speaker uses kam-question, he/she has already presented the presuppositions in his/her mind. Take (62) for example, the speaker expresses his attitude toward the topic of whether the man likes this woman. By using kam-questions, he/she has already shown that he does not believe the man will like the woman. (62) i. kam. he Q. 8. ga-i like. i?. (Taiwanese). her. Readers interested in kam-construction can refer to Lau (2010). 43.

(52) ‘Is it true that he like this girl?’ In addition, I have provide five pieces of evidence in section 2.4 to demonstrate the affinity between Hakka VP-mo questions and A-not-A questions. These similarities lead us to draw a conclusion that interrogative INFL is lexically realized in Hakka mo.. 2.6 Negation mang. As mentioned in the beginning of this chapter, mang generally shares the same syntactic distribution with mo. Though VP-mang is not a type of A-not-A questions, I would like to demonstrate that its derivation can still be accounted for by the NEG-to-I movement analysis. This indicates that our NEG-to-I movement is not restricted to VP-mo questions solely. In the following section, I demonstrate the syntactic distribution and semantic properties of mang first. Then the previous analysis of VP-mang will be introduced before I move to illustrate the derivation of VP-mang questions.. 2.6.1 Syntactic distribution and Semantic properties. Like Hakka mo, mang can appear immediately before the verb or at the end of the. 44.

(53) sentence. It can either serve as a sentential negation or a question marker9. Example (4a) and (4b) are repeated here as (63) and (64). (63) gi. mang. siid. fan.. he. NEG. eat. rice. ‘he hasn’t eaten yet.’ (64) gi. siid. fan mang?. he. eat. rice NEG. ‘Has he eaten yet?’ In terms of mang, it is hard to find a word-to-word correspondence in other Chinese dialects, such as Mandarin or Taiwanese Southern Min.. 9. Aside from preceding the verb and adding to the sentence final position so as to form a question, the negator. mang can be inserted into a compound word. Example (1b) illustrates that mang appears in the middle of the compound word siid ded ‘eat DED’. Interestingly, (1a) and (1b) share the same semantic meaning though the negation mang is located in different sentence positions. I leave this issue for further research (1). a.. ya-tiau. giung-zeu mang. siid ded.. this-CL. banana. eat. NEG-yet. DED. ‘This banana is not ripe enough that we can not eat.’ b.. ya-tiau. giung-zeu. siid mang. ded.. this-CL. banana. eat. DED. NEG-yet. ‘This banana is not ripe enough that we can not eat.’ c.. ya-tiau. giung-zeu sidd ded. mang?. this-CL. banana. NEG. eat. DED. ‘Is this banana ripe enough?’ 45.

(54) (65) ta he. hai-mei. chi. fan.. yet-not. eat. rice. ‘He hasn’t eaten yet’ (66) ta he. chi. fan. le-mei?. eat rich. Pref-NEG. ‘Has he eaten yet?’ Additional evidence comes from a comparison between Taiwanese Southern Min and Hakka. It is shown in (67) and (68) that Hakka mang matches iau beu ‘not yet’ in Taiwanese Southern Min. (67) a.. li. hit. pun tsa. you that CL. khaoN-liao. book read-finish. iau. bue?. yet NEG?. ‘Have you finished reading that book?’ b.. ngi. ge. bun. shu. kung-ted. mang?. you. that. CL. book. read-finish. yet-NEG. ‘Have you finished reading that book?’ (68) a.. i. iau bue. se. sing-ku.. he. yet-NEG. wash. body. He hasn’t taken a bath.. 46.

(55) b.. gi. mang. se. sinn?. he. yet-NEG. wash. body. ‘He hasn’t taken a bath.’. 2.6.2 Previous analysis on VP-mang. In dealing with VP-mang questions, Lai (2005) argued that the underlying structure of VP-mang is formed by an affirmative sentence and a negative sentence. These two sentences are connected by coordinator yahe ‘or’. VP-mang is derived by deleting various constituents within the affirmative and the negative sentence. I demonstrate the derivation of gi sii fang mang ‘ has he eaten yet’ step by step here. (69) a. Subject +you +VP(or NP)+e. b. Subject +mang you +VP (or NP).. gi you siid fan e. gi mang you siid fan. c. Subject +you +VP (or NP)+ e+ yahe+ Subject +mang you+ VP(or NP)? gi. you. siid. fan e. yahe. gi. mang. you. siid. fan. delete the second Subject d. Subject +you +VP (or NP)+ e+ yahe+ Subject +mang you+ VP(or NP)? gi. you. siid. fan e. yahe. gi. mang. you. siid. fan. delete the second word of the coordinator 47.

(56) and the repeated VP e. Subject +you +VP (or NP) +ya+ +mang you+? gi. you. siid. fan. ya. mang. you. delete the coordinator ‘ya’ and f.. ‘you’. Subject VP (or NP) mang? gi. siid. fan mang ? (adapted from Lai, 2005:181). I concur part of Lai’s analysis, but some problems make her analysis unattractive. It is possible to assume that the underlying structure of VP-mang is formed by an affirmative sentence and a negative sentence, but it is not reasonable to delete any part of the sentence without any theoretical basis. She does not elucidate the reason why some constitutes are deleted after another. Moreover, though she has mentioned that the negator mang in example (69f) moves directly from the pre-verbal position to the end of the sentence, she also does not clearly specify the intermediate or the final landing site of the negator mang. I attempt to provide a more elaborative derivation of VP-mang in my study.. 2.6.3 Deriving VP-mang questions. Since Hakka negation mang has the similar syntactic distribution with Hakka mo, I am 48.

(57) going to illustrate that our NEG-I-C movement can not only account for VP-mo questions but also VP-mang questions. The following two tree diagrams show the derivation of VP-mang questions. (70) Step one: NEG-I-C movment CP 3 C’ 3 C mangi. IP 3 gi I’ 3 I ti. NegP 3 Neg’ 3 Neg ti. VP 5 siid. (71) Step two: move the remaining IP to spec of C. CP 3 IPj 6 gi. siid. C’ 3 C mangi. fan. 49. tj. fan.

(58) 2.6 Summary. To sum up, in this chapter, I have dealt with two kinds of VP-NEG questions. For VP-mo questions, I argue that VP-mo questions are a type of A-not-A questions. Judging from its surface form, this argument seems not so convincing because there is no A-not-A form within VP-mo questions. However, I have shown that VP-mo questions behave like A-not-A questions in various ways. First, I demonstrated that VP-mo questions can not be answered by “Yes” or “No”; instead, the answer to the VP-mo questions needs to be the positive or negative form of the predicate. Secondly, VP-mo questions are not compatible with nandao’actually’ but daodi ‘truly’, which is one of the prominent characteristics of A-not-A questions. Third, like A-not-A questions, VP-mo questions are compatible with particles, such as ne and no in Mandarin Chinese and Hakka respectively. The last piece of evident is that VP-mo and A-not-A questions are both neutral questions. Speakers have no presupposition about the answer. Granted these similarities between VP-mo questions and A-not-A questions, I argue that mo is the overt realization of the Interrogative INFL. As for VP-mang questions, since the syntactic distribution of mang is similar to mo, I demonstrate that our NEG-I-C analysis can also apply to VP-mang questions as well.. 50.

(59) Chapter Three Classification of Hakka interrogatives. 3.1 Overview. Few studies, to the best of my knowledge, have been done on Hakka interrogatives (Luo,1985; Chung,2000; Wang,2008).Our analysis on VP-NEG questions has important implications for the classification of Hakka interrogatives. In particular, I argue that previous analyses on Hakka A-not-A questions and particle questions need to be modified. This chapter will be organized as follows: I begin by reviewing Chung’s work on the general overview of Hakka interrogatives. Then, differing from the previous classification, I will show that mo-interrogatives should not be viewed as particle questions in section 4.3. Moreover, I provide evidence to support my view that the apparent Hakka A-not-A questions as claimed in the literature actually belong to alternative questions in section 4.4. Besides, I discuss the inherent reason why [you......mo] construction noted in the previous literature can be treated as A-not-A questions. An interesting phenomenon will be demonstrated in section 4.5 as one piece of evidence for the existence of trace due to the application of ellipsis in various alternative questions. Finally I will conclude this chapter in section 4.6. 51.

(60) 3.2 Previous Classification of Hakka interrogatives. Like other Chinese dialects, Hakka. has several types of interrogative sentences,. including particle questions, alternative questions and wh-questions. Luo’s (1984) book on Hakka Grammar identified six types of interrogatives. Later in Chung’s (2000) work, he refines Luo’s classification and points out that there are four types of interrogatives in Hakka. I will discuss each type in turn. As displayed in (1) to (6), basically, Hakka wh-questions have the same syntactic property with Mandarin Chinese, in which they are formed by leaving a wh-word or wh-phrase in-situ rather than moving such an interrogative constituent to a clause-initial position as in English. (1) ng. hi. you go. si-cong. mai. ma-ge?. traditional market buy. what. ‘What did you buy in the traditional market?’ (2) ng you. hi. si-cong. du-do. ma-ngin?. go. traditional market. meet. who. ‘Who did you meet in the traditional market?’ (3) nai -kiu when. oi. hi. kuan a-gong?. will. go. see. grandpa. ‘When do we go to visit grandpa?’ 52.

(61) (4) iong-e how. zhong. zo-ted. da. koi?. adv. can. bite. open. ‘How can I open this?’ (5) gi he. zuo-ma-ge. oi. gio-zhi?. why. want. cry. ‘Why does he cry?’ (6) ng he-do he live. nai-vi? where. ‘Where does he live?’ The second type of Hakka interrogatives is called particle questions. It is constructed by simply adding a particle to the sentence-final position of the declarative sentences. Compared with Mandarin Chinese, Hakka has more than twenty kinds of sentence final particles (Li, 2006; Cheng, 2007). In Chung’s work, he briefly mentioned that different question particles will have different pragmatic functions as illustrated in example (7). Furthermore, he pointed out that VP-mo questions belong to particle questions. Specifically, for him, Hakka mo is on a par with Mandarin ma. (7) gi he. zo-ded. hi. gung-ien. gao. mo/ho/ka?. can. go. park. play. particle. ‘Can he go to the park?’ 53.

(62) Examples (8) and (9) illustrate characteristics of alternatives questions. In alternative questions, there will be two or more than two choices for the listener to choose from. And those options will be combined with a coordinator jahe. (8) ng. oi. You want. siid fang. jehe. siid miang?. eat rice. or. eat noodle. ‘Which one would you want to eat, rice or noodles?’ (9) ng you. hi-fon. cong-go-e. tiau-vu. jahe. vag-tu. like. sing. dance. or. draw. ‘Which one do you like, sing, dance or draw?’ Last, Chung mentioned that the canonical example of Hakka A-not-A interrogative sentences is formed by a positive element disjoined with its negative counterpart with the ya morpheme, as exemplified in (10) to (12). (10) ng. fuang-xi. You. happy. ya or. m fuang-xi? not. happy. ‘Do you feel happy?’ (11) ya this. liang shangfu ngi CL. clothes. you. oi. mai. ya. want. buy. or not. “Do you want to buy this clothe?”. 54. mo. oi. mai?. want. buy?.

(63) (12) ya. bun. this CL. shu. ngi. zhong-i. book. you. like. ya or. m. zhong-i?. not. like?. ‘Do you like this book?’. 3.3 Genuine Particles. In this section, I am going to show that sentences with mo in the final position should not be treated on par with Mandarin ma. Let us consider Mandarin ma questions first, as shown in examples (13) and (14), the particle ma can be attached to both affirmative and negative statements. (13). nainai. xihuan. zhe-ge. nuhai ma?. grandma. like. this-CL girl. particle. ‘Does grandma like this girl?’ (14). nainai. bu. grandma not. xihuan. zhe-ge. nuhai. like. this-CL girl. ma? particle?. ‘Doesn’t grandma like this girl?’ However, a sharp contrast can be observed in Hakka mo questions. Hakka mo is not fully grammaticalized to a pure particle just as ma in Mandarin Chinese. Hence, unlike particle questions, mo-interrogatives require the predicate to be affirmative, as represented in (15). 55.

(64) Example (16) demonstrates that mo is unable to be used in a negative sentence. But, if we change negation mo into a genuine particle ho, it would turn an ungrammatical sentence into a grammatical sentence. (15) apo grandma. zhongi. ya-ge. se-moi-e. like. this-CL girl. mo? NEG?. ‘Does grandma like this girl?’ (16) apo. m. grandma not. se –moi-e. zhongi. ya-ge. like. this-CL girl. *mo/ho? *NEG/particle?. ‘Doesn’t grandma like this girl’. 3.4Genuine A-not-A questions. A few scholars approach Hakka A-not-A questions in a descriptive fashion (Chung, 2000; Tsai, 2002; Wang, 2008). In this section, I would like to argue against previous works on viewing two types of construction as A-not-A questions.. 3.4.1 X-ya-m-X≠A-not-A. I argue that it is problematic to view X-ya-m-X questions as a type of A-not-A 56.

(65) questions in Hakka. I will use two tests proposed by Huang (1991) and MaCawly (1994) respectively to test the X-ya-m-X interrogative questions. Huang (1991) pointed out that the true A-not-A questions exhibit systematic island properties with respect to their distribution and interpretation, whereas haishi-questions are free from island constraints. Now let’s consider the Mandarin Chinese data first and then examine the Hakka data. In Mandarin Chinese, haishi questions can occur within the island, such as sentential subject islands or relative clause islands, without inducing any island violation. In (17), we have hasishi-questions within the sentential subject and in (18) we have haishi-questions inside the relative clauses. All of them are grammatical sentences. (17) [wo qu I. go. Meiguo haishi American or. bu. qu. Meiguo ]. bijiao. hao?. not go. American. more. good?. ‘Is it better that I go to American or that I do not go to American?’ (18) ni you. xihuang [renshi. ni. like. you. know. haishi bu or. renshi. ni]. not know. you. de. ren?. DE person?. ‘Do you like people who know you or people who don’t know you?’ Nevertheless, if we locate A-not-A questions inside the island, this will yield ungrammatical sentences, as shown in (19) and (20). (19) *[wo. qu. bu. qu. I. go. not go. Meiguo] bijiao hao? American more good? 57.

(66) ‘Is it better that I go to American or not?’ (20) *ni You. xihuan. [renshi. bu. renshi. ni. de. ren]?. like. know. not. know. you. DE person?. ‘Do you like people who know you or don’t know you?’ Since we know that haishi-questions differ from true A-not-A questions in their sensitivity to island constraints. Thus, I am going to show that X-ya-m-X is not a true A-not-A questions due to the reason that it is not constrained by the two islands noted above. The following examples demonstrate X-ya-m-X can appear in sentential subject islands and relative clause islands , the same as haishi-quesiton in Mandarin Chinese. (21) gi. hi. ya. m. hi. miguo. ka. ho?. he. go. or. not. go. American. more. good. ‘Is it better that I go to American or that I do not go to American?’ (22) ngi You. hifon. ngica. ya. m. ngica. ngi. ge. nin?. like. know. or. not know. you. DE. person?. ‘Do you like people who know you or don’t know you?’ Another test comes from McCawley (1994), who observed that when positive and negative elements are conjoined by haishi, the order of these two conjuncts is free. However, a true A-not-A question strictly requires A to occur before Not A. (23). Ta. daodi. lai. (haishi) bu 58. lai?.

(67) He truly. come. (or). not come. ‘Let me get to the answer: will he come or not?’ (24). Ta. daodi. He truly. bu. lai. *(haishi) lai?. not come. or. come. ‘Let me get to the answer: will he come or not?’ Again, based on MaCawley’s observation, it is shown that Hakka X-ya-m-X questions can reorder the two elements. (25) ngi. cidou. ya. dong. vog-e. jiong. ya. m. jiong?. you. think. this. CL. house. pretty. or. not pretty. ‘Do you think this is a big house or not? (26) ngi You. cidou. ya. dong. vug-e. m. think. this. CL. house. not. jiong pretty. ya. jiong?. or. pretty. ‘Do you think this is a big house or not?’ (27) ngi You. xifon. ya. m. xifon. like. or. not like. ya. bu. tianyang?. this. CL. movie. ‘Do you like this movie or not?’ (28) ngi You. m. xifon. not like. ya. xifon. ya. bu. tiangyang?. or. like. this. CL. movie. ‘Do you like this movie or not?’ 59.

(68) Following my analysis, it is clear that the examples discussed in the previous works behave like alternative questions. That is because ya serves as a coordinator in connecting a positive constituent with its negative counterpart. Questions with a overt disjunctive coordinator should not be seen as genuine A-not-A questions.. 3.4.2 you......mo≠A…… not-A. When negation marker mo is used as a question marker, it is usually paired with its affirmative counterpart you, as shown in (29) to (30). Under this circumstance, Tsai (2002) pointed out that in Hakka, as A-not-A questions are concerned, the affirmative part (A) and the negative part (not-A) are separated. Specifically, the negative part (not-A) does not immediately follow the affirmative part (A); instead, the negative part is located in the sentence final position. (29) gi he. you. zi. vug-ga. mo?. have. at. home. NEG. you. oi. hi. mo?. have. want. go. NEG. ‘Is he at home?’ (30) ng you. ‘Do you want to go?’ 60.

(69) I agree with part of Tsai’s argument in which examples (29) and (30) can be regarded as A-not-A questions, but the reason why (29) and (30) can be treated as A-not-A questions is because mo is the overt realization of Interrogative INFL not because you serves as positive element (A) and mo serves as negative element (not-A). The paradigm in (31) illustrates that in Mandarin, A-not-A questions should include both a positive element (A) and a negative element (not A). Absence of either A part or not-A part will convert the A-not-A question into a declarative sentence, as demonstrated in (31b) and (31c). (31) a. ta he. xihuan. bu. xihuan. like. not like. zhe. ge. nvsheng?. this. CL girl.. (A-not-A). ‘Does he like this girl or doesn’t he like this girl?’ b.. ta. xihuan. zhe ge. nvsheng.. he. like. this CL girl. ‘He likes this girl.’ c.. ta. bu. xihuan. he. not like. zhe ge. nvsheng.. this CL girl. ‘He doesn’t like this girl.’ Now, let us consider Hakka [you……mo] construction. If you serves as A part, the 61.

(70) absence of you will convert the question into a declarative sentence. However, examples (32) and (33) show that it is not obligatory for you to appear in [you……mo] construction.. The. absence of you does not either render the sentence ungrammatical or make the original sentence loss the meaning of question. (32)gi he. (you). zi. vug-ga. mo?. have. at. home. NEG. ‘Is he at home?’ (33) ng. (you). you have. oi. hi. mo?. want. go. NEG. ‘Do you want to go?’. 3.5 Tond Sandhi. The typical disjunctive questions may include a paradigm like (34). (34) a.. ta. xihuan na-ben. he. like. shu. haishi. that-CL book. or. bu. xihuan. not like. na-ben. that-CL book. ‘Does he like this book or doesn’t he like this book?’ b.. ta. xihuan. naben. shu. he. like. that-CL book. haishi or 62. bu. xihuan [e]?. not like. shu?.

(71) ‘Does he like that book or doesn’t (he) like (that book)?’ c.. ta. xihuan. he. like. [e] haishi or. bu. xihuan. not like. naben. shu?. that-CL book. ‘Does he like (that book) or doesn’t (he ) not like that book?’ No matter under Wang’s one-rule approach or Huang’s Modular approach, researchers have reached a consensus on the fact that the entire paradigm can be derived by suitable deletion10 rule .The deletion of zhe-ben-shu ‘ this book’ in both (34b) and (34c) will leave traces. This section provides a piece of evidence for supporting the existence of the trace. Intriguingly, it is observed that tone sandhi will be blocked by the trace in the alternative questions in Hakka. Hakka tone sandhi is repeated here again. Yinping 24 undergoes tone sandhi when it is followed by yingping 24, qusheng 55 and yangru 5. (35). Hakka Tone Sandhi: 24. 24+. 55 5. 10. 24 11 +. 55 5. I use the term deletion, since in terms of example (36b), Huang favors anaphoric ellipsis while Li (2002). argues that this might be a kind of VP ellipsis. Without devoting myself to a particular analysis, the main point here is to show that AB-not-A form is derived from AB-not-AB by applying the process of deletion. And the trace will be left in the original position 63.

(72) Now, let us consider some crucial empirical data with regard to alternative data in Hakka. Supposedly, since the coordinator ya is qusheng 55, any word of yingping 24 preceding it will change to yangping 11, as demonstrated in (36) and (37). (36). mai. ya. m. mai. Buy. or. not. buy. Base tone. 24. 55. 11. 24. Tone sandhi. 11. 55. 11. 24. qiam. ya. m. qiam. sign. or. not. sign. Base tone. 24. 55. 11. 24. Tone sandhi. 11. 55. 11. 24. (37). Nevertheless, tone sandhi will not apply to the same constituent when it appears in the alternative sentence. I argue that the reason why tone sandhi does not occur is because there is a trace left by the application of ellipsis between the verb and the coordinator. Thus, this empty category prohibits the process of tone sandhi. (38) ng You. oi. mai [e]. want. buy. ya mo voi or. not want. mai. ya-lian san-fu?. buy. his-CL clothes. ‘Do you want to buy this clothes or you don’t want to buy this clothes?’ (39) gi. voi qiam [e] ya. m. voi qiam 64. ya-fun. kei-iog?.

(73) he. will sign. or. not will. sigh. this-CL contract. ‘Will he sign this contract or will he not sign this contract?’ (40). mai. [e] ya. m. mai. Base tone. 24. 55 11. 24. Tone sandhi. 24. 55. 11. 24. m. qiam. (41). qiam [e]. ya. Base tone. 24. 55. 11. 24. Tone sandhi. 24. 55. 11. 24. 3.6 Summary In the beginning of this chapter, I first review the previous classification of Hakka interrogative. Later, it is pointed out that two kinds of questions should be re-evaluated. I argue that VP-mo questions do not belong to particle questions by showing that VP-mo questions can not co-exist with the negator m. Next, we claim that X-ya-m-X should not be regarded as Hakka A-not-A questions but as alternative questions for the reason that X-ya-m-X is free from island constraints and the order of the positive and negative element can be exchanged. Also, I elucidate the inherent reason why [you……mo] construction would be treated as A-not-A questions. Finally, I showed that the process of tone sandhi becomes inactivated due to the trace left by the process of ellipsis in alternative questions. 65.